Llm on Pensées de Michel

Sandogasa, now for EPEL and CentOS SIG workflows too!

Thu, 02 Apr 2026 00:00:00 +0000

Caveat lector

This post discusses tools reluctantly written with AI assistance. If you don’t entertain using them under any circumstance, and think even reading about them legally compromise your ability to reimplement them yourselves, stop reading now

This is a follow-up to the original Sandogasa announcement. Before I ended up fedora-cve-triage to extract library crates and reuse them in the other Sandogasa tools, I already created two tools for managing CentOS Hyperscale SIG workflows, hs-intake and hs-relmon. It simply makes sense to also merge them back in and deduplicate functionalities.

The D&D alignments of developers w.r.t. AI

Wed, 01 Apr 2026 00:00:00 +0000

I have a more serious post coming, but I figured better not risk being confused for a bad joke (especially given the chardet controversy.

So instead - something I’ve been promising some people I would write about … the idea that we can classify developers based on their attitude towards (generative) AI, a.k.a. LLMs.

Mind you, don’t take this too seriously, and especially, don’t feel singled out. I’m making fun of myself too here! Suggestions for improvements welcome, of course.

Announcing Sandogasa, a set of Rust crates to help with Fedora packaging workflows

Tue, 24 Mar 2026 19:06:34 +0000

Caveat lector

This post discusses tools reluctantly written with AI assistance. If you don’t entertain using them under any circumstance, and think even reading about them legally compromise your ability to reimplement them yourselves, stop reading now

I’ve spent the past few weeks of having to use LLMs to scratch some long-standing itches that, unfortunately, no one in the community has had the time to solve programmatically.

fedora-cve-triage

It started off with fedora-cve-triage, written to address the issue that a lot of CVE bugs filed against Fedora packages are badly attributed, and there is a lack of automation for handling issues filed against CVEs that have been addressed in a software update but failed to reference said issue.

Hi, my name is Michel, and I (reluctantly) use LLMs

Tue, 24 Mar 2026 15:41:31 +0000

the old pen resists
but the cursor blinks, waiting—
I press Enter. Fine.

~ Claude Opus 4.6 (1M context)

If you know where I work, you’ve probably heard of news reports that we will be judged on AI-driven impact. I’ll let you drawn your own conclusion on how much truth there is in the reports, but you can listen to what Zuck said about AI in a recent earnings report.